Understand the difference between IaaS and PaaS and know when each SQL platform is used.
Why Should I Learn This?
Imagine you have designed a database.
The next question is:
"Where will I run this database?"
Microsoft offers several SQL platforms, and each platform provides a different type of control and responsibility.
This blog is not about writing SQL queries. It is about understanding where SQL Server runs and who manages it.
First, Understand Cloud Computing
Suppose you want to open a restaurant. There are two choices.
Option 1 — Build Everything Yourself
You buy:
- Land
- Building
- Kitchen
- Electricity
- Security
- Furniture
You maintain everything yourself. This is similar to:
Infrastructure as a Service (IaaS)
You have maximum control. You also have maximum responsibility.
Option 2 — Rent a Fully Managed Restaurant
The owner provides:
- Building
- Electricity
- Security
- Maintenance
You only cook food. This is similar to:
Platform as a Service (PaaS)
You focus on your application instead of infrastructure.
Understanding the Diagram
The diagram has two axes.
Horizontal Axis
On-Premises ------------------------> Cloud
(More Control) (Less Administration)
As you move to the right:
- Microsoft manages more.
- You manage less.
Vertical Axis
Higher Cost
↑
|
|
Lower Cost
Generally:
- More dedicated resources = Higher cost
- More shared cloud resources = Lower cost
Platform 1 — SQL Server (On-Premises)
Your Company
↓
Own Physical Server
↓
Install Windows
↓
Install SQL Server
↓
Create Database
Who manages everything?
You.
Responsibilities:
- Buy hardware
- Install operating system
- Install SQL Server
- Configure networking
- Configure security
- Backup database
- Patch SQL Server
- Replace failed hardware
Advantages:
- Maximum control
- Full customization
Disadvantages:
- Expensive
- Time-consuming
- Requires database administrators (DBAs)
Platform 2 — SQL Server on Azure Virtual Machine (IaaS)
Microsoft gives you:
- A virtual machine
You install:
- Windows or Linux
- SQL Server
Responsibilities:
✔ Configure SQL Server
✔ Patch SQL Server
✔ Configure backups
✔ Manage security
✔ Performance tuning
Microsoft manages
- Physical servers
- Azure networking
- Storage
- Data center
Think of this as:
Renting a computer instead of buying one.
Platform 3 — Azure SQL Database (PaaS)
Microsoft already provides:
- SQL Server
- Operating System
- Updates
- Automatic Backups
- High Availability
- Hardware
- Networking
You only create:
- Databases
- Tables
- Indexes
- Views
- Stored Procedures
- Constraints
- Data
Your focus becomes:
Database Design
instead of
Server Administration
Platform 4 — Azure SQL Managed Instance
Imagine your company already has a large SQL Server application. It uses advanced SQL Server features. Moving directly to Azure SQL Database may require application changes.
Azure SQL Managed Instance solves this problem.
It provides:
- Almost complete SQL Server compatibility
while Microsoft still manages:
- Infrastructure
- Updates
- Backups
- High Availability
Think of it as:
SQL Server in the cloud with minimal application changes.
Comparing IaaS and PaaS
| Feature | IaaS | PaaS |
|---|---|---|
| Physical Server | Microsoft | Microsoft |
| Virtual Machine | You | Microsoft |
| Operating System | You | Microsoft |
| SQL Server Installation | You | Microsoft |
| SQL Updates | You | Microsoft |
| Backups | You | Microsoft |
| Database Design | You | You |
| Tables | You | You |
| Views | You | You |
| Stored Procedures | You | You |
| Data | You | You |
Easy Way to Remember
IaaS
"I manage the server."
PaaS
"I manage the database."
Real-Life Example
Imagine you work as a database developer.
Your manager says:
"Don't waste time installing Windows or patching SQL Server.
Just design the database."
Which platform should you choose?
Answer: Azure SQL Database (PaaS)
DP-800 Exam Tips
"Which platform reduces infrastructure management?"
Answer: Azure SQL Database (PaaS)"Which platform gives maximum control over SQL Server?"
Answer: SQL Server on Azure Virtual Machine (IaaS)
Summary
- SQL Server (On-Premises) → You manage everything.
- SQL Server on Azure VM → Microsoft manages hardware, you manage the VM and SQL Server.
- Azure SQL Database → Microsoft manages the infrastructure; you focus on database development.
- Azure SQL Managed Instance → Similar to Azure SQL Database, but designed for easier migration of existing SQL Server applications.
Key Takeaways
- ✔ IaaS = More control, more administration.
- ✔ PaaS = Less administration, more focus on database development.
- ✔ Microsoft manages the infrastructure in PaaS.
- ✔ Developers mainly work with tables, indexes, views, stored procedures, and data.
Now moving forward.........................
"The Hyperscale service tier eliminates many of the practical limitations traditionally associated with cloud databases..."
A beginner like me immediately asks:
- What is a service tier?
- What is Hyperscale?
- What is a node?
- What is storage architecture?
- What is a replica?
- What is a read-intensive workload?
Microsoft explains Hyperscale, but assumes you already know the other five concepts.
Before We Begin. Remember this:
A database has two parts.
Part 1 : Infrastructure
- Hardware
- Network
- Operating System
- SQL Server installation
- Updates
- Backups
-Part 2 : Database
- Tables
- Views
- Indexes
- Stored Procedures
- Data
The difference between Microsoft's SQL platforms is simply:
Who manages Part 1?
Azure SQL Database
What is Azure SQL Database?
- Azure SQL Database is Microsoft's fully managed cloud database
- You do NOT install SQL Server
- You do NOT patch Windows.
- You do NOT replace failed hardware.
Microsoft does all of that. You simply create databases and write SQL.
Think of it as: Google Docs instead of Installing Microsoft Word.
What does Fully Managed mean?
Microsoft manages:
✔ Hardware
✔ Network
✔ Operating System
✔ SQL Server Updates
✔ Automatic Backup
✔ High Availability
You manage
✔ Tables
✔ Data
✔ Views
✔ Stored Procedures
✔ Indexes
What is a Service Tier?
Imagine buying a mobile internet package.
Basic Package
↓
Professional Package
↓
Unlimited Package
Azure SQL Database also has different plans. These are called
Service Tiers
Each tier gives different performance.
Hyperscale
Microsoft writes
Hyperscale removes storage limits.
Let's understand. Suppose your database grows like this.
10 GB
↓
100 GB
↓
2 TB
↓
20 TB
Many traditional databases eventually reach storage or performance limits. Hyperscale is designed to keep growing without needing to redesign your database.
Benefits
✔ Storage automatically grows
✔ No predefined maximum size
✔ Pay only for what you use
✔ Better performance for very large databases
What is a Replica?
Imagine 1,000 people want to read the same book. One copy of the book becomes crowded. Instead, the library creates five copies. Now everyone can read simultaneously.
Those extra copies are called Replicas They mainly help with Read Operations
Read Workload
SELECT
Reading data.
Write Workload
INSERT
UPDATE
DELETE
Changing data.
Read-intensive workload
Most users are reading data rather than changing it.
Example News website: Thousands of people reading Very few people editing.
Serverless
Suppose nobody connects to your database.
Traditional Server
Running
↓
Consumes CPU
↓
Costs Money
Serverless
Nobody connected?
↓
Automatically Pause
↓
No Compute Charges
↓
Someone connects
↓
Automatically Resume
This reduces costs.
Why does Microsoft mention Retry Logic?
When a paused database wakes up, it needs a few seconds. Your application should automatically retry the connection instead of immediately showing an error.
Automatic Tuning
Normally, a DBA analyzes slow queries. Creates indexes. Improves performance.
Azure SQL Database can do much of this automatically.
Think of it as
Database
↓
Detect Slow Query
↓
Recommend Index
↓
Sometimes Create It Automatically
High Availability
Suppose one server fails. Microsoft automatically switches your database to another server. Users usually don't notice.Microsoft guarantees 99.99% availability.
Azure SQL Managed Instance
Instead of saying
Near 100% compatibility...
Imagine your company already has a SQL Server application.
It uses
- SQL Server Agent
- Linked Servers
- Database Mail
- Service Broker
Moving directly to Azure SQL Database may require code changes. Azure SQL Managed Instance supports most SQL Server features. It lets companies move to Azure with minimal application changes.
SQL Server on Azure VM
Think of Azure VM as renting a computer.
Microsoft gives you:
✔ Virtual Machine
You install:
✔ Windows
✔ SQL Server
✔ Configure Everything
Maximum Control
Maximum Responsibility
SQL Database in Microsoft Fabric
This is where most tutorials fail.
I'd explain it like this:
Traditional Company
Application
↓
SQL Database
↓
ETL
↓
Data Warehouse
↓
Power BI
Many separate systems.
Fabric
Application
↓
SQL Database
↓
Automatic Mirroring
↓
OneLake
↓
Analytics
↓
Power BI
No manual ETL. Everything stays synchronized automatically.
OneLake
Think of OneLake as Google Drive for enterprise data. Every Fabric service shares the same storage.
Delta Parquet
This is NOT another database. It is a highly optimized file format used for analytics. Fabric automatically creates these files from your SQL tables.
Why?
Analytics queries can be extremely heavy. Instead of querying the live production database, Fabric queries the Delta Parquet copy. Your application remains fast.
Summary Table
| Platform | Best Choice When |
|---|---|
| SQL Server | Full control on your own servers |
| SQL Server on Azure VM | Need OS/SQL customization |
| Azure SQL Database | Building modern cloud applications |
| Azure SQL Managed Instance | Migrating existing SQL Server apps |
| SQL Database in Fabric | Need OLTP + analytics + AI together |
Reference : understanding-platform-choices
Top comments (0)